EmarationRemediate in the order that retires the most legal risk first.
Public K-12 districts and community colleges face thousands of pages and tiny IT teams. This estimator ranks your accessibility findings by enforcement exposure — how likely a defect is to trigger an OCR complaint or DOJ action, weighted by who it blocks, where it lives, and how widespread it is — and pairs each with a defensible remediation-hours range.
It ranks exposure; it does not issue a "compliant" verdict. The output is a dated, good-faith measurement as of the report date — never legal advice.
Two pure engines, grounded in the case record
Enforcement-risk ranking
Every finding is scored 0–100 and tiered (critical / high / moderate / low) from a fully decomposed formula: enforcement frequency, assistive-tech impact, content scope, and reach, scaled by the most severe WCAG level and bumped for DOJ-named barriers. The enforcement ordering — PDFs > captions > alt text > forms/keyboard > contrast/color-only > LMS — is the observed frequency order in the OCR/DOJ record, not a guess.
Remediation-hours estimate
Counted backlogs become a defensible effort range (min / typical / max), priced per PDF page and per video-minute where that — not the document count — drives the work. Every assumption (pages per PDF, minutes per video, the volume-efficiency curve) is returned so a human reviewer can adjust it per engagement.
How the score is built
exposure = 0.35·enforcement + 0.25·userImpact + 0.25·scope + 0.15·reach
score(0–100) = round( 100 · clamp01( exposure · legalSeverity + dojBarrierBump ) )
Tiers: score ≥ 75 critical · ≥ 55 high · ≥ 35 moderate · else low. Every weight is a documented, tunable constant, and every assessment carries its factor breakdown plus a sourced rationale — so a reviewer and counsel can audit each number.
What it will not do
Never fabricate
An unknown finding type or invalid count is refused, not guessed. Every finding cites a real WCAG success criterion with provenance and detector confidence.
Flag exceptions, never apply them
A possible 28 CFR 35.201 exception is recorded for a human/counsel determination — it does not silently change the score or exclude a finding.
Counted ≠ passing
PDFs, media, and manual-required checks are reported as "not yet assessed", never as "passing". Low detector confidence flags verification — it never down-ranks a real risk.